Exploring patterns in European singles charts

Andrzej Buda1,2, Andrzej Jarynowski3,4

1Wydawnictwo Niezależne, Wrocław, Poland2 IFJ, Polish Academy of Science, Cracow, Poland

3Jagiellonian University, Institute of Physics, Cracow, Poland4Moldova State University, Kishinev, R. of Moldova

andrzej.jarynowski@sociology.su.se, andrzejbudda@gmail.com

Abstract. European singles charts are important part of the music industry responsible for creating popularityof songs. After modeling and exploring dynamics of global album sales in previous papers, we investigatepatterns of hit singles popularity according to all data (1966-2015) from weekly charts (polls) in 12 WesternEuropean countries. The dynamics of building popularity in various national charts is more than the economybecause it depends on spread of information. In our research we have shown how countries may be affectedby their neighbourhood and influenced by technological era. We have also computed correlations withgeographical and cultural distances between countries in analog, digital and Internet era. We have shown thattime delay between the single premiere and the peak of popularity has become shorter under the influence oftechnology and the popularity of songs depends on geographical distances in analog (1966-1987) and Internet(2004-2015) era. On the other hand, cultural distances between nations have influenced the peaks ofpopularity, but in the Compact Disc era only (1988-2003). We have also indicated the European countries “inline” with global trends e.g. The Netherlands, the United Kingdom and outsiders like Italy and Spain.

Keywords: computational social science, data-mining of social behavior, information diffusion

Introduction

In academic studies, as well as real-world settings (business-oriented or fan-oriented) many metrics ofmeasuring popularity [1] and distance were proposed [2]. Computer science and mathematics introduced a plentyof ranking algorithms as Google’s PageRank of WWW sites and its variation.

Music – mediated by sounds – is widely enjoyed by people all over the planet, across a diverse background ofnumerous cultures. Some common bonds of music are dynamics, texture, rhythm and pitch. While music haveaccompanied humans on their journeys through life, and the people themselves, in turn, the created of themusical trends and established businesses to supply demand on music.

Recently, the knowledge of computational tools for sociology and music such as recommendations [3] hasundergone an accelerating growth, however all models of such system are incomplete without real data,especially register-based [4]. The requirement to calibrate the algorithms encourages cooperation betweenvarious registering institutions, which, in turn, exerts a pressure on collecting data for simple analysis by manyresearchers who work on new models and use complex tools often taken from other disciplines [5]. We presentcomputational approach to this issue. We based our project on different approach from beginning withmathematical though computer science to finish with sociological and statistical analyses. This work is also veryspecious, because of its interdisciplinary and showing light on methods which were not use in “music studies”till this moment. Only a few researches were provided for understanding successes or failures of an artist, mainlyin economic frameworks of product life-cycle or innovation spread. There were also culture studies focused onethnology and customers decisions. Unfortunately, quantitative methods are rarely used in these analysis. If it isapplied, linear regression dominates. A slightly less well known but related superstar phenomenon is even moreinteresting from complex system perspective due to herd effect and criticality of customers decision. Music hashad an impact on society for as long as people have made music and listened to it. Music has acted as a meansfor giving a cultural voice to people – consumers. Music even settles communities: for instance, punk, techno,and ethnic music all embody some type of subculture that encompasses an entire lifestyle. In this work, wemodel the challenges for artists and their labels staying behind their market decisions in ultimate sales game. Theproperties of phonographic market have been recently explored from economic point of view including thestochastic model of global phonographic market and the influence of digital piracy on albums sales dynamics[6]. Like in financial markets, according to album sales history, it is possible to use Minimum Spanning Treesmethods (MST) and distinguish sectors and subsectors of artists that represent various music genres. However,that kind of hierarchical structure does not contain sectors that represent pop music, because there are two groups

of customers: music fans that buy records every week, and occasional clients who spend their money on musicbefore Christmas or Valentine's Day [7]. This second group is only interested in celebrities from various musicgenres and makes record sales grow up at the end of each year. According to these economic reasons, pop musicdoes not exist in hierarchical structure of the market [6], but in the Minimum Spanning Tree we have the sectorof the most famous celebrities that represent various genres. Thus, the group of occasional customers isresponsible for seasonality that affects the global dynamics of albums sales.

On the other hand, there is also another part of phonographic market - the hit singles. Since Thomas Edison'sphonograph, the short cuts of music has always been sold each year in form of 7" vinyl singles, 3" or 5" CDsingles, mp3, ringtones, radio or TV broadcast, etc. The variety of forms is much bigger that physical long-playing records format, so the dynamics of hit singles popularity should be considered from informationtheory/spread and computational sciences point of view. Because each national singles chart is based on differentrules. For example, the UK hit singles chart depends on physical or digital sales only, but the Italian singles chartis a combination of singles sales and airplays from the local radio and TV. The national singles markets havevarious sizes, so the sales results are usually published weekly. However, in the sixties and seventies the singlescharts in Austria and France have been published monthly and the West German chart has been published bi-weekly.

There has been a lot of research into epidemiologic or opinions spread models [8], especially based ondiffusions [9] or interactions between two nearest neighbours (the Ising model). The dynamics of hit singlespopularity has such properties because one European nation might affect or have been affected by others in theirneighbourhood.

Diffusion of information and influence spread within cultural sphere are also topics of broad evaluation. Wehave studied the diffusion of popularity of singles [10] in Western Europe in order to compare time delays oftopping in the national charts. Countries are constantly influencing their surroundings and being influenced byothers. The topological characteristics of networks consequently determine dynamical processes on top of thenetwork, e.g. cascade of information adoption or default contagion in culture networks [11-14]. It gives us anopportunity to observe the spread of music popularity.

Data analysis and background information

In our research we have collected all the hit singles data from the last 50 years. We have selected the mostpopular singles in the period 1966-2015 that have reached the weekly published Top 15 in the national hitsingles charts including: 01 AUSTRIA (A) 02 BELGIUM (B) 03 SWITZERLAND (CH) 04 GERMANY (D) 05SPAIN (E) 06 FRANCE (F) 07 GREAT BRITAIN (GB) 08 ITALY (I) 09 IRELAND (IRL) 10 NORWAY (N)11 NETHERLANDS (NL) 12 SWEDEN (S)

These data have been published weekly on European music papers, radio or TV broadcast and the Internet forthe last 50 years. Of course, since 1966 we have had the physical analog formats of music (7 inch singles, 12inch maxi singles), digital 3 or 5 inch singles that have been dominated the charts in the nineties (1988-2001)and the streaming of official digital mp3, ringtones (2002 until now). We have also investigated how the spreadof hit singles through various European countries may depend on these formats of music dominating variousages.

In our research we have considered trajectories of a hit single performed on national charts as a productlifecycle with the highest position on each chart. If the highest position belongs to the Top 15, the country hasbeen affected. Otherwise, the single has not been a hit, so the country is not affected. As a main data parameter twe consider a number of weeks (after initial single release) when the single reaches the peak of the nationalrecord chart. If the single does not affect the country (or has a peak outside the Top 15), we set the value of thisparameter t = 52 weeks, because in phonography the maximum length of seasonal popularity is usually limited toone year.

We have analysed all period (1966-2015) where initial singles releases have been more or less contagious andcome from various countries. Usually, the hit singles popularity have come from the neighbourhood countries.According to our results, the United Kingdom, the Netherlands and Belgium have ability to be affected fast, aftera few weeks (Fig. 1). On the other hand, Spain and Italy have been affected after t = 20 weeks after initial hitsingle release. This kind of behaviour suggests geographical dependences, but in our research we focus onvarious eras (1966-1987, 1988-2003, 2004-2015) and consider also cultural dependences between Europeancountries.

Fig. 1 The map of western Europe according to spread of music popularity for hit singles (1966-2015)

We investigate distances between European countries in Cartesian / Euclidean two dimensional meaning. Themost typical representation are geographical distances. However, many other matrixes of distances are possible.For example cultural map of Europe was created by political scientists based on the World Values Survey [15].Two dominant dimensions were selected (explaining 70% of variations between countries): traditional versussecular-rational values on the vertical y-axis and survival versus self-expression values on the horizontal x-axis(Fig. 2).

Fig. 2 The map of western Europe according to geographic and cultural distance. Links are based on flowsbetween counties (inverse gravity force)

In physical and digital era (LP, CD) the product life-cycle had its maximum a few weeks after the release,because music labels need more time to supply physical phonograms and satisfy the number of recordsdemanded by customers. On the other hand, in recent Internet era, t usually tends to 1, because records mayreach the peak of popularity in the first week after the release. The example of this behavior in global recordsales trajectories of Pink Floyd (1979) and Coldplay (2007) is given by (Fig. 3).

Fig. 3 The position of The Wall by Pink Floyd (1979, upper line) and Viva La Vida by Coldplay (2007) onglobal albums sales chart in consecutive weeks after the release.

However, the patterns in local European singles charts mean more than just the economy, because allcountries might be affected by their geographical neighbourhood (Fig 1). Moreover, the cultural distances havealso influenced on information spread, and peaks of popularity. Thus, we have decided to show normalizeddistances between countries according to geography, culture and t - time delay between peaks of popularity aftersingles premieres (Fig. 4). The double cross on the third diagram (Fig. 3) reflects the fact that Spain and Italy hasbecome the most isolated countries according to spread of hit singles popularity in Europe.

Fig. 4 Normalized distances between countries (in geography, culture and music popularity)

Results – Phases.

In our research we explore patterns in European singles chart in long time period (1966-2015) where the timedelay t between record premiere and the peak of popularity has become shorter and shorter. This phenomenarefers to singles and albums (Fig. 2). So it is reasonable to distinguish analog (1966-1987), digital (1988-2003)and Internet (2004-2015) era in record industry, because of the technology.

Fig. 5 Boxplots of distrubuitions of delays in peeaking top position on charts in analog, digital and Internet era.

1) Analog era (1966-1987)

In the early '50s of the 20th century, as a physical format, the 7 inch singles were the simpliest and basicformats of songs that make record industry able to compile national record sales charts. We have observedgeographical dependences between popularity spreads. Of course, most of hit singles have initial roots in the UKand gained global popularity quickly, but we also considered songs from all 12 European countries that hadnational charts on that time. If the hit single came from distant countries like Norway or Sweden, it took half orone year to affect all Europe. For example, Take On Me by Norwegian trio A-ha gained the peak of popularityand effected all 12 European countries in the late 1985, one year after initial release. Thus, the correlationbetween t and geographical distance [Tab. 1] is observed for all countries (Fig. 5).

2) Digital era (1988-2003)

The '80s of the 20th century introduced the „undestructible” digitality to music industry in physical forms andCompact Discs dominated phonographic markets all over the world. In digital the peak of national charts hadbeen reached almost equally in all countries. Moreover, we observe significant correlations between t andcultural distances in Europe [Fig. 4, Tab. 1]. On the other hand, the end of 20 th century brought almost allphonographic markets into European integration and unity. So, even the Italy, that had never created aninternational hit in 7” singles analog era (1966-1987), had been finally able to initialize the popularity ofDragostea Din Tei by Moldovian band O-Zone (2003). The language of this song was understandable toimigrants from Romania that had been able to move and settle in other European countries in the early 21 th

century. The dependence between music popularity and geographical distances disappeared in that era [Tab. 1].

3) Internet Era (2004-2015).

The growing power of Internet in the beginning of 21th century caused dramatic decrease of albums sales allover the world [16] because people could have direct access to music files by Internet connection. There was noneed to collect music in traditional physical formats like CDs or vinyls because of mp3s and ringtones. On the

other hand, music labels developped streaming technologies to stop digital piracy. Thus, all national charts hadbecome compiled also from digital singles sales. The Internet era caused globalization in the phonographicmarkets all over the world (2004-2015). Thus, the time t of reaching peaks of popularity had become muchshorter than before and correlations between t and culture distances had disappeared. On the other hand, wecould observe weak correlations between t and geographical distances [Tab. 1] because all effects of the Internetera mean stronger dependence on information spread.

Fig. 6 Hierarchical representations of delays t in peaking top position on charts in analog, digital an Internet era

Subdominant ultrametric space that provides hierarchy of dependence in other perspective like diagrams (Fig.6). This method describes collective relations between elements [17] – countries in Europe. We have alsocomputed the Manhattan distances between peaks of popularity in 12-dimensional space represented by 12national charts. Thus, it is possible to detect hierarchical structure of European singles chart in ultrametric space.It has been necessary to show the evolution in hierarchical structure of European countries. The United Kingdomhas always been a leader in spreading popularity of songs. The most distant country from the rest is Italy,because a large number of European hit singles (including Dancing Queen by Abba or Macarena by Los DelRio) has never gained popularity there. However, this phenomena of isolation has disappeared completelybecause of the Internet era. On the other hand, another distant country – Spain – has always been affected latebecause of geographical reasons. Moreover, Spain is going to be more isolated than before, because a number of21st century popular hits has never reached Spanish charts even in the Internet era.

Tab. 1 Correlations between networks of distances between European countries. Music delays categories 1-Analog, 2- Digital, 3- Internet, where med - median and mean – mean delays. Geo – geographical and cult –cultural distance stand for reference.

Conclusions and future works.

Diversity of gaining popularity in European countries is significant and has become strongly investigated inthe Internet era in spite of globalization [18]. The whole history of record industry shows that music labels tendto supply brand new records to customers as fast as it is possible. Thus, in the next consecutive technological era,time delays t between the hit single release and the height of popularity become shorter and shorter (Fig. 5, 6).Thus, the network and information society has been driven by the technological forces. However, eachtechnological era has its own patterns with different possibility of modelling it [19]:

- Analog regime: huge variety between countries and medium variety within countries (Fig. 5), positivespatial correlations [Tab. 1]. The nationality of all considered singles charts in Europe has meaning.Geography plays a main role in spread of popularity (grid modelling approach could describe that erawell).

- Digital regime: huge variety inside the countries, but small differences between countries (Fig. 5),positive correlations with culture distances [Tab. 1]. It could be the effect of European integration andunification (decreasing distance between countries). The variety of t within each country is much bigger,

probably because of new music genres. Here, cultural similarities have the strongest impact oninformation spread. To model this system in proper way, networks approach is needed.

- Internet era: medium variety between countries and medium variety within country (Fig. 5), nocorrelations significant [Tab. 1]. Despite of globalization, local nationalism come back on the stage. Here,modelling would be more difficult, because none of presented topologies (Fig. 2) may describeinformation spread well.

Results of our visualization analysis may develop new (more reliable) dynamic description music popularityspread patterns. Some temporal [20] as well as spatial [21] relations were already known to scientific com-munity, but not from singles populararity perspective. However, our results are limited to 12 available Europeansingles charts, so the effect of isolation in Spanish singles charts in the Internet era remains unexplained. If thisphenomena is caused by globalisation, it might be probably successfully explained by considering data and influ-ences from other non-European countries including Latin America. On the other hand, from the non-Europeanpoint of view, the leadership of the United Kingdom might be the reminiscence of the United States and otherEnglish-speaking countries. To answer these questions, we are going to apply agent-based modelling to globalphonographic market in our future works [19].

References

[1] Caves, R. E. Creative industries: Contracts between art and commerce. Harvard University Press, 2000.[2] Jajuga, K. and Walesiak, M. Remarks on the dependence measures and the distance measures. PRACE

NAUKOWE AKADEMII EKONOMICZNEJ WE WROCŁAWIU. 1022, 2004.[3] Jannach, D., et al. What Recommenders Recommend – An Analysis of Accuracy, Popularity, and Sales

Diversity Effects. User Modeling, Adaptation, and Personalization. pp 25-37, 2013.[4]. Grabowski, A, Kruszewska, N and Kosiński, R. Properties of on-line social systems.. 2008, Eur. Phys.

J. B, Vol. 66, pp. 107-113.[5] Jarynowski, A. Obliczeniowe nauki społeczne w praktyce. Wrocław : WN, 2014.[6] Buda, A. and Jarynowski, A. Lifetime of Correlation and its applications. Wroclaw : WN, 2011.[7] Buda, A. and Jarynowski, A. Network structure of phonographic market with characteristic similarities

between artists. Acta Physica Polonica A , Vol. 123. 2013 [8] Jankowski, J., Michalski, R. and Kazienko, P. The Multidimensional Study of Viral Campaigns as

Branching Processes. 2012, LNCS, Vol. 7710, pp. 462-474[9] Goel, S., Watts, D. J. and Goldstein, D. G The Structure of Online Diffusion Networks. 2012, ACM

Journal, Vol.[10] Ordanini, A. 2006, Journal of Cultural Economics, Vol. 30, pp. 183-200.[11] Dybiec, B, Namiko, M and Sneppen, K. Information spreading and development of cultural centers.

Physical Review E. 2012, Vol. 85, 5.[12] Rogers, E.M. Diffusion of Innovations. New York : Free Press, 2005.[13] Nowak, A., Szamrej, J. and Latané, B. From private attitude to public opinion: A dynamic theory of

social impact. 1990, Psychological Review, Vol. 97, pp. 362-376.[14] Grabowski, A. The relationship between human behavior and the process of epidemic spreading in a

real social network. 2012, EPJ B, Vol. 85, p. 248.[15] Inglehart, R. Welzel, C, Modernization, Cultural Change and Democracy, New York and Cambridge:

Cambridge University Press. 2005.[16] IFPI. The IFPI Report Recording industry in numbers, 2012 Edition. London, 2013.[17] Buda, A, and Jarynowski A. "Durability of Links between Assets in Financial Markets: Minimal

Spanning Trees and Correlations." 2014 European Network Intelligence Conference (ENIC). IEEE, 2014.[18] Castells, M. The rise of the network society: The information age: Economy, society, and culture. Vol. 1.

John Wiley & Sons, 2011.[19] Buda, A, and Jarynowski A. Agent-based modeling, complex networks and system dynamics –

practical approaches, Полтава: ПУЕТ, Інформатика та системні науки (ІСН-2015): матеріали VIВсеукраїнської науково-практичної конференції, 2015.

[20] Percino, G., Klimek, P., & Thurner, S. Instrumentational complexity of music genres and whysimplicity sells. PloS one, 2014, 9(12), e115255.

[21] Lee, C., & Cunningham, P. The geographic flow of music. In Proceedings of the 2012 InternationalConference on Advances in Social Networks Analysis and Mining (ASONAM 2012) (pp. 691-695). IEEEComputer Society.

Appendix

Tab. 2 List of the 50 most popular singles in 1966-2015:Analog Digital Internet

SCOTT MCKENZIE - SAN FRANCISCOEDDY GRANT - GIMME HOPEJO'ANNA

CRAZY FROG - AXEL F

THE BEATLES - HEY JUDE MARIAH CAREY - WITHOUT YOUSHAKIRA/WYCLEFF JEAN - HIPSDON'T LIE

SERGE GAINSBOURG/JANE BIRKIN - JE T'AIME...MOINON PLUS

SCATMAN JOHN - SCATMAN GNARLS BARKLEY - CRAZY

MUNGO JERRY - IN THE SUMMERTIME LOS DEL RIO - MACARENA RIHANNA/JAY-Z - UMBRELLA

GEORGE HARRISON - MY SWEET LORDELTON JOHN - CANDLE IN THEWIND

TIMBALAND/ONE REPUBLIC -APOLOGIZE

VICKY LEANDROS - APRES TOICELINE DION - MY HEART WILLGO ON

DUFFY - MERCY

ROLLING STONES - ANGIE CHER - BELIEVE BEYONCE - IF I WERE A BOYGEORGE MCCRAE - ROCK YOUR BABY LOU BEGA - MAMBO NO.5 LADY GAGA - POKER FACEGEORGE BAKER SELECTION - PALOMA BLANCA BOMFUNK MC'S - FREESTYLER ADELE - ROLLING IN THE DEEP

ABBA - DANCING QUEENKYLIE MINOGUE - CAN'T GET YOUOUT OF MY HEAD

RIHANNA/CALVIN HARRIS - WEFOUND LOVE

BONEY M - MA BAKERSHAKIRA - WHENEVERWHEREVER

PSY - GANGNAM STYLE

J.TRAVOLTA/OLIVIA N.JOHN - YOU'RE THE ONE THAT IWANT

LAS KETCHUP - ASEREJEDAFT PUNK/PHARRELL WILLIAMS -GET LUCKY

VILLAGE PEOPLE - YMCA EMINEM - LOSE YOURSELF PHARRELL WILLIAMS - HAPPY

PINK FLOYD - ANOTHER BRICK IN THE WALL O-ZONE - DRAGOSTEA DIN TEILILLYWOOD/THE PRICK - PRAYERINC.

BARBRA STREISAND - WOMAN IN LOVEMARK RONSON/BRUNO MARS -UPTOWN FUNK

PAUL MCCARTNEY/STEVIE WONDER - EBONY ANDIVORYDAVID BOWIE - LET'S DANCESTEVIE WONDER - I JUST CALLED TO SAY I LOVE YOUA.HA - TAKE ON MEEUROPE - THE FINAL COUNTDOWNMICHAEL JACKSON - I JUST CAN'T STOP LOVING YOU